Letta

Active
GitHub Python Apache-2.0

Description

Letta (formerly MemGPT) is an open-source framework for building stateful AI agents with advanced reasoning and transparent long-term memory. It allows you to visually test, debug, and observe agents.

Key Features

  • Persistent long-term memory — agents remember user preferences, context, and history across sessions
  • Memory block management — structured state via configurable memory blocks (human, persona)
  • CLI interface — run memory-powered agents directly in terminal via Letta Code CLI
  • REST API integration — Python/TypeScript SDK to embed stateful agents into your apps
  • Skills and subagent support — built-in skill system and subagent orchestration
  • Model-agnostic architecture — works with any LLM backend, freely switchable

Use Cases

💡 Building long-term personal assistants that remember user preferences
💡 Developing customer service agents requiring cross-session context
💡 Creating self-improving coding assistants with learning capabilities
💡 Building multi-step workflow agents needing state management
💡 Creating enterprise AI applications with transparent memory mechanisms

Quick Start

npm install @letta-ai/letta-client

import Letta from "@letta-ai/letta-client";

const client = new Letta({ apiKey: process.env.LETTA_API_KEY });

const agentState = await client.agents.create({
  model: "openai/gpt-5.2",
  memory_blocks: [
    { label: "human", value: "Name: Alice. Role: developer." },
    { label: "persona", value: "I am a helpful coding assistant." }
  ],
  tools: ["web_search", "fetch_webpage"],
});

const response = await client.agents.messages.create(agentState.id, {
  input: "What do you know about me?",
});

for (const message of response.messages) {
  console.log(message);
}

Related Projects

Related Articles